Development of a Web navigation guide system based on the hypertext probabilistic grammar

Finding some relevant information from a large collection of data on a particular Web site is not an easy task, since many users are frequently faced with a huge number of alternative links that may be followed. Hence, it would be useful for these users if they were provided with a tool that can assist them in performing the Web navigation processes. This paper explains the development of a Web navigation guide system capable of giving users a set of recommended Web links on a Web site based on a set of rules obtained from a Web log miner; i.e. a Web usage mining application aimed at exploiting user navigation rules from a set of users´ navigation sessions recorded in a Web access log file. Navigation sessions representing user interactions with several Web sites are modeled as a hypertext probabilistic grammar. In this model, since strings or navigation rules having high probability reflect navigation paths that are preferred by users for some reasons, they are used by the system to determine a set of recommended links to be followed. The system, which is deployed as an additional pop-up menu on the browser, is implemented using the Oracle8i database as well as PHP4 scripting language and Java script on the server and client sides, respectively. In addition to this, apart from its main function for providing users with a set of recommended navigation paths, the Web guide navigation system might also be used to monitor the user navigation positions.